In two previous studies, we observed that recombinant human interleukin- 3 (IL-3) induced an increase in marrow burst-forming unit-erythroid- derived colonies in vitro in some patients with Diamond-Blackfan anemia (DBA). To determine whether a similar erythropoietic response could be induced in vivo, we treated 13 patients with DBA (aged 4 to 19 years) with two preparations of IL-3. All patients had absent absolute reticulocyte counts and markedly reduced to absent recognizable bone marrow erythroid elements; patients with circulating reticulocytes in the previous 12 months were excluded from study. All patients except 1 had failed steroid therapy and had been transfusion-dependent since infancy; 1 patient was maintained on high-dose prednisone at the time of enrollment. On the first arm of the study, IL-3 (Immunex Corp, Seattle, WA) was administered subcutaneously using a dose escalation regimen of 125 to 500 micrograms/m2/day in divided dosage at 12-hour intervals, coadministered with 1.5 mg/kg/d of oral ferrous sulphate. Of the 13 patients that entered the trial, 4 stopped prematurely because of adverse side effects. In the other 9 evaluable cases, reticulocytes increased transiently in 1 patient from 0 to 65 x 10(9)/L after 35 days of IL-3 therapy at 250 micrograms/m2, but transfusion dependency persisted. One transient peak in absolute reticulocyte count was noted in 6 other patients, but no erythroid response was observed after completion of a full course of IL-3. Oral prednisone at 0.5 mg/kg/d was then coadministered with IL-3 at 500 micrograms/m2 to 5 of the patients without effect, and treatment was stopped. In 2 patients, a second preparation of IL-3 (Sandoz Canada Inc, Dorval, Quebec, Canada) was initiated in a dose escalation regimen of 2.5 to 10 micrograms/kg and was coadministered with ferrous sulphate. No erythroid response was observed in either patient, and in one of the two, alternate-day subcutaneous recombinant erythropoietin at 300 U/kg was administered for 3 weeks in combination with daily IL-3 at 10 micrograms/kg, but no increased erythropoiesis was seen. Significant increases in white blood cell and eosinophil counts during administration of both preparations of IL-3 were observed in all patients. These data show that the response of DBA patients to IL-3 in vivo is heterogeneous and cannot be predicted from in vitro studies. The absence of a corrective effect of IL-3 in these patients with DBA indicates that a deficiency of the cytokine is not central in the pathogenesis of the disorder.